3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
28 #include "BufferEncodings.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "output_docbook.h"
41 #include "ParagraphParameters.h"
42 #include "SpellChecker.h"
44 #include "texstream.h"
45 #include "TextClass.h"
48 #include "WordLangTuple.h"
51 #include "frontends/alert.h"
53 #include "insets/InsetBibitem.h"
54 #include "insets/InsetLabel.h"
55 #include "insets/InsetSpecialChar.h"
56 #include "insets/InsetText.h"
58 #include "mathed/InsetMathHull.h"
60 #include "support/debug.h"
61 #include "support/docstring_list.h"
62 #include "support/ExceptionMessage.h"
63 #include "support/gettext.h"
64 #include "support/lassert.h"
65 #include "support/lstrings.h"
66 #include "support/textutils.h"
73 using namespace lyx::support;
75 // OSX clang, gcc < 4.8.0, and msvc < 2015 do not support C++11 thread_local
76 #if defined(__APPLE__) || (defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ < 8)
77 #define THREAD_LOCAL_STATIC static __thread
78 #elif defined(_MSC_VER) && (_MSC_VER < 1900)
79 #define THREAD_LOCAL_STATIC static __declspec(thread)
81 #define THREAD_LOCAL_STATIC thread_local static
88 /// Inset identifier (above 0x10ffff, for ucs-4)
89 char_type const META_INSET = 0x200001;
94 /////////////////////////////////////////////////////////////////////
98 /////////////////////////////////////////////////////////////////////
100 class SpellResultRange {
102 SpellResultRange(FontSpan range, SpellChecker::Result result)
103 : range_(range), result_(result)
106 FontSpan const & range() const { return range_; }
108 void range(FontSpan const & r) { range_ = r; }
110 SpellChecker::Result result() const { return result_; }
112 void result(SpellChecker::Result r) { result_ = r; }
114 bool contains(pos_type pos) const { return range_.contains(pos); }
116 bool covered(FontSpan const & r) const
118 // 1. first of new range inside current range or
119 // 2. last of new range inside current range or
120 // 3. first of current range inside new range or
121 // 4. last of current range inside new range
122 //FIXME: is this the same as !range_.intersect(r).empty() ?
123 return range_.contains(r.first) || range_.contains(r.last) ||
124 r.contains(range_.first) || r.contains(range_.last);
127 void shift(pos_type pos, int offset)
129 if (range_.first > pos) {
130 range_.first += offset;
131 range_.last += offset;
132 } else if (range_.last >= pos) {
133 range_.last += offset;
138 SpellChecker::Result result_ ;
142 /////////////////////////////////////////////////////////////////////
146 /////////////////////////////////////////////////////////////////////
148 class SpellCheckerState {
152 needs_refresh_ = true;
153 current_change_number_ = 0;
156 void setRange(FontSpan const & fp, SpellChecker::Result state)
159 RangesIterator et = ranges_.end();
160 RangesIterator it = ranges_.begin();
161 for (; it != et; ++it) {
162 if (!it->covered(fp))
163 result.push_back(SpellResultRange(it->range(), it->result()));
164 else if (state == SpellChecker::WORD_OK) {
165 // trim or split the current misspelled range
166 // store misspelled ranges only
167 FontSpan range = it->range();
168 if (fp.first > range.first) {
169 // misspelled area in front of WORD_OK
170 range.last = fp.first - 1;
171 result.push_back(SpellResultRange(range, it->result()));
174 if (fp.last < range.last) {
175 // misspelled area after WORD_OK range
176 range.first = fp.last + 1;
177 result.push_back(SpellResultRange(range, it->result()));
182 if (state != SpellChecker::WORD_OK)
183 ranges_.push_back(SpellResultRange(fp, state));
186 void increasePosAfterPos(pos_type pos)
188 correctRangesAfterPos(pos, 1);
192 void decreasePosAfterPos(pos_type pos)
194 correctRangesAfterPos(pos, -1);
198 void refreshLast(pos_type pos)
200 if (pos < refresh_.last)
204 SpellChecker::Result getState(pos_type pos) const
206 SpellChecker::Result result = SpellChecker::WORD_OK;
207 RangesIterator et = ranges_.end();
208 RangesIterator it = ranges_.begin();
209 for (; it != et; ++it) {
210 if(it->contains(pos)) {
217 FontSpan const & getRange(pos_type pos) const
219 /// empty span to indicate mismatch
220 static FontSpan empty_;
221 RangesIterator et = ranges_.end();
222 RangesIterator it = ranges_.begin();
223 for (; it != et; ++it) {
224 if(it->contains(pos)) {
231 bool needsRefresh() const
233 return needs_refresh_;
236 SpellChecker::ChangeNumber currentChangeNumber() const
238 return current_change_number_;
241 void refreshRange(pos_type & first, pos_type & last) const
243 first = refresh_.first;
244 last = refresh_.last;
247 void needsRefresh(pos_type pos)
249 if (needs_refresh_ && pos != -1) {
250 if (pos < refresh_.first)
251 refresh_.first = pos;
252 if (pos > refresh_.last)
254 } else if (pos != -1) {
255 // init request check for neighbour positions too
256 refresh_.first = pos > 0 ? pos - 1 : 0;
257 // no need for special end of paragraph check
258 refresh_.last = pos + 1;
260 needs_refresh_ = pos != -1;
263 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
265 needs_refresh_ = true;
268 current_change_number_ = change_number;
271 typedef vector<SpellResultRange> Ranges;
272 typedef Ranges::const_iterator RangesIterator;
274 /// the area of the paragraph with pending spell check
277 /// spell state cache version number
278 SpellChecker::ChangeNumber current_change_number_;
281 void correctRangesAfterPos(pos_type pos, int offset)
283 RangesIterator et = ranges_.end();
284 Ranges::iterator it = ranges_.begin();
285 for (; it != et; ++it) {
286 it->shift(pos, offset);
292 /////////////////////////////////////////////////////////////////////
294 // Paragraph::Private
296 /////////////////////////////////////////////////////////////////////
298 class Paragraph::Private
300 // Enforce our own "copy" constructor
301 Private(Private const &) = delete;
302 Private & operator=(Private const &) = delete;
303 // Unique ID generator
304 static int make_id();
307 Private(Paragraph * owner, Layout const & layout);
308 /// "Copy constructor"
309 Private(Private const &, Paragraph * owner);
310 /// Copy constructor from \p beg to \p end
311 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
314 void insertChar(pos_type pos, char_type c, Change const & change);
316 /// Output the surrogate pair formed by \p c and \p next to \p os.
317 /// \return the number of characters written.
318 int latexSurrogatePair(BufferParams const &, otexstream & os,
319 char_type c, char_type next,
320 OutputParams const &);
322 /// Output a space in appropriate formatting (or a surrogate pair
323 /// if the next character is a combining character).
324 /// \return whether a surrogate pair was output.
325 bool simpleTeXBlanks(BufferParams const &,
326 OutputParams const &,
329 unsigned int & column,
331 Layout const & style);
333 /// This could go to ParagraphParameters if we want to.
334 int startTeXParParams(BufferParams const &, otexstream &,
335 OutputParams const &) const;
337 /// This could go to ParagraphParameters if we want to.
338 bool endTeXParParams(BufferParams const &, otexstream &,
339 OutputParams const &) const;
342 void latexInset(BufferParams const &,
347 Font const & outerfont,
349 Change & running_change,
350 Layout const & style,
352 unsigned int & column);
355 void latexSpecialChar(
357 BufferParams const & bparams,
358 OutputParams const & runparams,
359 Font const & running_font,
360 string & alien_script,
361 Layout const & style,
364 unsigned int & column);
371 unsigned int & column);
377 unsigned int & column);
383 unsigned int & column);
386 void validate(LaTeXFeatures & features) const;
388 /// Checks if the paragraph contains only text and no inset or font change.
389 bool onlyText(Buffer const & buf, Font const & outerfont,
390 pos_type initial) const;
392 /// a vector of speller skip positions
393 typedef vector<FontSpan> SkipPositions;
394 typedef SkipPositions::const_iterator SkipPositionsIterator;
396 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
398 Language * getSpellLanguage(pos_type const from) const;
400 Language * locateSpellRange(pos_type & from, pos_type & to,
401 SkipPositions & skips) const;
403 bool hasSpellerChange() const
405 SpellChecker::ChangeNumber speller_change_number = 0;
406 if (theSpellChecker())
407 speller_change_number = theSpellChecker()->changeNumber();
408 return speller_change_number > speller_state_.currentChangeNumber();
411 bool ignoreWord(docstring const & word) const ;
413 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
415 pos_type textsize = owner_->size();
416 // check for sane arguments
417 if (to <= from || from >= textsize)
419 FontSpan fp = FontSpan(from, to - 1);
420 speller_state_.setRange(fp, state);
423 void requestSpellCheck(pos_type pos)
426 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
428 speller_state_.needsRefresh(pos);
431 void readySpellCheck()
433 speller_state_.needsRefresh(-1);
436 bool needsSpellCheck() const
438 return speller_state_.needsRefresh();
441 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
443 speller_state_.refreshRange(first, last);
445 last = owner_->size();
448 pos_type endpos = last;
449 owner_->locateWord(first, endpos, WHOLE_WORD, true);
452 owner_->locateWord(last, endpos, WHOLE_WORD, true);
457 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
461 while (it != et && it->first < start) {
462 int skip = it->last - it->first + 1;
470 void markMisspelledWords(pos_type const & first, pos_type const & last,
471 SpellChecker::Result result,
472 docstring const & word,
473 SkipPositions const & skips);
475 InsetCode ownerCode() const
477 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
480 /// Which Paragraph owns us?
484 Inset const * inset_owner_;
493 ParagraphParameters params_;
495 /// for recording and looking up changes
499 InsetList insetlist_;
502 pos_type begin_of_body_;
504 typedef docstring TextContainer;
508 typedef set<docstring> Words;
509 typedef map<string, Words> LangWordsMap;
513 Layout const * layout_;
515 SpellCheckerState speller_state_;
519 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
520 : owner_(owner), inset_owner_(nullptr), id_(-1), begin_of_body_(0), layout_(&layout)
527 int Paragraph::Private::make_id()
529 // The id is unique per session across buffers because it is used in
530 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
533 static atomic_uint next_id(0);
538 Paragraph::Private::Private(Private const & p, Paragraph * owner)
539 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
541 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
542 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
545 requestSpellCheck(p.text_.size());
549 Paragraph::Private::Private(Private const & p, Paragraph * owner,
550 pos_type beg, pos_type end)
551 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
552 params_(p.params_), changes_(p.changes_),
553 insetlist_(p.insetlist_, beg, end),
554 begin_of_body_(p.begin_of_body_), words_(p.words_),
557 if (beg >= pos_type(p.text_.size()))
559 text_ = p.text_.substr(beg, end - beg);
561 FontList::const_iterator fcit = fontlist_.begin();
562 FontList::const_iterator fend = fontlist_.end();
563 for (; fcit != fend; ++fcit) {
564 if (fcit->pos() < beg)
566 if (fcit->pos() >= end) {
567 // Add last entry in the fontlist_.
568 fontlist_.set(text_.size() - 1, fcit->font());
571 // Add a new entry in the fontlist_.
572 fontlist_.set(fcit->pos() - beg, fcit->font());
574 requestSpellCheck(p.text_.size());
578 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
579 bool output_active, TocBackend & backend) const
581 d->changes_.addToToc(cdit, buf, output_active, backend);
585 bool Paragraph::isDeleted(pos_type start, pos_type end) const
587 LASSERT(start >= 0 && start <= size(), return false);
588 LASSERT(end > start && end <= size() + 1, return false);
590 return d->changes_.isDeleted(start, end);
594 bool Paragraph::isChanged(pos_type start, pos_type end) const
596 LASSERT(start >= 0 && start <= size(), return false);
597 LASSERT(end > start && end <= size() + 1, return false);
599 return d->changes_.isChanged(start, end);
602 // FIXME: Ideally the diverse isChanged() methods should account for that!
603 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
605 LASSERT(start >= 0 && start <= size(), return false);
606 LASSERT(end > start && end <= size() + 1, return false);
608 for (auto const & icit : d->insetlist_) {
609 if (icit.pos < start)
613 if (icit.inset && icit.inset->isChanged())
619 bool Paragraph::isChanged() const
621 return d->changes_.isChanged();
625 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
627 // keep the logic here in sync with the logic of eraseChars()
631 Change const & change = d->changes_.lookup(size());
632 return change.inserted() && change.currentAuthor();
635 Change Paragraph::parEndChange() const
637 return d->changes_.lookup(size());
641 void Paragraph::setChange(Change const & change)
643 // beware of the imaginary end-of-par character!
644 d->changes_.set(change, 0, size() + 1);
647 * Propagate the change recursively - but not in case of DELETED!
649 * Imagine that your co-author makes changes in an existing inset. He
650 * sends your document to you and you come to the conclusion that the
651 * inset should go completely. If you erase it, LyX must not delete all
652 * text within the inset. Otherwise, the change tracked insertions of
653 * your co-author get lost and there is no way to restore them later.
655 * Conclusion: An inset's content should remain untouched if you delete it
658 if (!change.deleted()) {
659 for (pos_type pos = 0; pos < size(); ++pos) {
660 if (Inset * inset = getInset(pos))
661 inset->setChange(change);
667 void Paragraph::setChange(pos_type pos, Change const & change)
669 LASSERT(pos >= 0 && pos <= size(), return);
670 d->changes_.set(change, pos);
672 // see comment in setChange(Change const &) above
673 if (!change.deleted() && pos < size())
674 if (Inset * inset = getInset(pos))
675 inset->setChange(change);
679 Change const & Paragraph::lookupChange(pos_type pos) const
681 LBUFERR(pos >= 0 && pos <= size());
682 return d->changes_.lookup(pos);
686 void Paragraph::acceptChanges(pos_type start, pos_type end)
688 LASSERT(start >= 0 && start <= size(), return);
689 LASSERT(end > start && end <= size() + 1, return);
691 for (pos_type pos = start; pos < end; ++pos) {
692 switch (lookupChange(pos).type) {
693 case Change::UNCHANGED:
694 // accept changes in nested inset
695 if (Inset * inset = getInset(pos))
696 inset->acceptChanges();
699 case Change::INSERTED:
700 d->changes_.set(Change(Change::UNCHANGED), pos);
701 // also accept changes in nested inset
702 if (Inset * inset = getInset(pos))
703 inset->acceptChanges();
706 case Change::DELETED:
707 // Suppress access to non-existent
708 // "end-of-paragraph char"
710 eraseChar(pos, false);
721 void Paragraph::rejectChanges(pos_type start, pos_type end)
723 LASSERT(start >= 0 && start <= size(), return);
724 LASSERT(end > start && end <= size() + 1, return);
726 for (pos_type pos = start; pos < end; ++pos) {
727 switch (lookupChange(pos).type) {
728 case Change::UNCHANGED:
729 // reject changes in nested inset
730 if (Inset * inset = getInset(pos))
731 inset->rejectChanges();
734 case Change::INSERTED:
735 // Suppress access to non-existent
736 // "end-of-paragraph char"
738 eraseChar(pos, false);
744 case Change::DELETED:
745 d->changes_.set(Change(Change::UNCHANGED), pos);
747 // Do NOT reject changes within a deleted inset!
748 // There may be insertions of a co-author inside of it!
756 void Paragraph::Private::insertChar(pos_type pos, char_type c,
757 Change const & change)
759 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
762 changes_.insert(change, pos);
764 // This is actually very common when parsing buffers (and
765 // maybe inserting ascii text)
766 if (pos == pos_type(text_.size())) {
767 // when appending characters, no need to update tables
769 // but we want spell checking
770 requestSpellCheck(pos);
774 text_.insert(text_.begin() + pos, c);
776 // Update the font table.
777 fontlist_.increasePosAfterPos(pos);
780 insetlist_.increasePosAfterPos(pos);
782 // Update list of misspelled positions
783 speller_state_.increasePosAfterPos(pos);
787 bool Paragraph::insertInset(pos_type pos, Inset * inset,
788 Font const & font, Change const & change)
790 LASSERT(inset, return false);
791 LASSERT(pos >= 0 && pos <= size(), return false);
793 // Paragraph::insertInset() can be used in cut/copy/paste operation where
794 // d->inset_owner_ is not set yet.
795 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
798 d->insertChar(pos, META_INSET, change);
799 LASSERT(d->text_[pos] == META_INSET, return false);
801 // Add a new entry in the insetlist_.
802 d->insetlist_.insert(inset, pos);
804 // Some insets require run of spell checker
805 requestSpellCheck(pos);
811 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
813 LASSERT(pos >= 0 && pos <= size(), return false);
815 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
818 Change change = d->changes_.lookup(pos);
820 // set the character to DELETED if
821 // a) it was previously unchanged or
822 // b) it was inserted by a co-author
824 if (!change.changed() ||
825 (change.inserted() && !change.currentAuthor())) {
826 setChange(pos, Change(Change::DELETED));
827 // request run of spell checker
828 requestSpellCheck(pos);
832 if (change.deleted())
836 // Don't physically access the imaginary end-of-paragraph character.
837 // eraseChar() can only mark it as DELETED. A physical deletion of
838 // end-of-par must be handled externally.
844 d->changes_.erase(pos);
846 // if it is an inset, delete the inset entry
847 if (d->text_[pos] == META_INSET)
848 d->insetlist_.erase(pos);
850 d->text_.erase(d->text_.begin() + pos);
852 // Update the fontlist_
853 d->fontlist_.erase(pos);
855 // Update the insetlist_
856 d->insetlist_.decreasePosAfterPos(pos);
858 // Update list of misspelled positions
859 d->speller_state_.decreasePosAfterPos(pos);
860 d->speller_state_.refreshLast(size());
866 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
868 LASSERT(start >= 0 && start <= size(), return 0);
869 LASSERT(end >= start && end <= size() + 1, return 0);
872 for (pos_type count = end - start; count; --count) {
873 if (!eraseChar(i, trackChanges))
879 // Handle combining characters
880 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
881 otexstream & os, char_type c, char_type next,
882 OutputParams const & runparams)
884 // Writing next here may circumvent a possible font change between
885 // c and next. Since next is only output if it forms a surrogate pair
886 // with c we can ignore this:
887 // A font change inside a surrogate pair does not make sense and is
888 // hopefully impossible to input.
889 // FIXME: change tracking
890 // Is this correct WRT change tracking?
891 Encoding const & encoding = *(runparams.encoding);
892 docstring latex1 = encoding.latexChar(next).first;
893 if (runparams.inIPA) {
894 string const tipashortcut = Encodings::TIPAShortcut(next);
895 if (!tipashortcut.empty()) {
896 latex1 = from_ascii(tipashortcut);
899 docstring latex2 = encoding.latexChar(c).first;
901 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
902 // Encoding supports the combination:
903 // output as is (combining char after base char).
904 os << latex2 << latex1;
905 return latex1.length() + latex2.length();
908 os << latex1 << "{" << latex2 << "}";
909 return latex1.length() + latex2.length() + 2;
913 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
914 OutputParams const & runparams,
917 unsigned int & column,
919 Layout const & style)
921 if (style.pass_thru || runparams.pass_thru)
924 if (i + 1 < int(text_.size())) {
925 char_type next = text_[i + 1];
926 if (Encodings::isCombiningChar(next)) {
927 // This space has an accent, so we must always output it.
928 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
933 if (runparams.linelen > 0
934 && column > runparams.linelen
936 && text_[i - 1] != ' '
937 && (i + 1 < int(text_.size()))
938 // same in FreeSpacing mode
939 && !owner_->isFreeSpacing()
940 // In typewriter mode, we want to avoid
941 // ! . ? : at the end of a line
942 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
943 && (text_[i - 1] == '.'
944 || text_[i - 1] == '?'
945 || text_[i - 1] == ':'
946 || text_[i - 1] == '!'))) {
948 os.texrow().start(owner_->id(), i + 1);
950 } else if (style.free_spacing) {
959 void Paragraph::Private::latexInset(BufferParams const & bparams,
961 OutputParams & runparams,
964 Font const & outerfont,
966 Change & running_change,
967 Layout const & style,
969 unsigned int & column)
971 Inset * inset = owner_->getInset(i);
974 if (style.pass_thru) {
975 odocstringstream ods;
976 inset->plaintext(ods, runparams);
981 // FIXME: move this to InsetNewline::latex
982 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
983 // newlines are handled differently here than
984 // the default in simpleTeXSpecialChars().
985 if (!style.newline_allowed) {
989 bool needPar = false;
990 column += running_font.latexWriteEndChanges(
991 os, bparams, runparams,
992 basefont, basefont, needPar);
996 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
999 basefont = owner_->getLayoutFont(bparams, outerfont);
1000 running_font = basefont;
1002 if (runparams.moving_arg)
1006 os.texrow().start(owner_->id(), i + 1);
1010 if (owner_->isDeleted(i)) {
1011 if( ++runparams.inDeletedInset == 1)
1012 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1015 if (inset->canTrackChanges()) {
1016 column += Changes::latexMarkChange(os, bparams, running_change,
1017 Change(Change::UNCHANGED), runparams);
1018 running_change = Change(Change::UNCHANGED);
1022 odocstream::pos_type const len = os.os().tellp();
1024 if (inset->forceLTR(runparams)
1025 && running_font.isRightToLeft()
1026 // ERT is an exception, it should be output with no
1027 // decorations at all
1028 && inset->lyxCode() != ERT_CODE) {
1029 if (runparams.use_polyglossia) {
1031 } else if (running_font.language()->lang() == "farsi"
1032 || running_font.language()->lang() == "arabic_arabi")
1033 os << "\\textLR{" << termcmd;
1039 // FIXME: Bug: we can have an empty font change here!
1040 // if there has just been a font change, we are going to close it
1041 // right now, which means stupid latex code like \textsf{}. AFAIK,
1042 // this does not harm dvi output. A minor bug, thus (JMarc)
1044 // Some insets cannot be inside a font change command.
1045 // However, even such insets *can* be placed in \L or \R
1046 // or their equivalents (for RTL language switches), so we don't
1047 // close the language in those cases.
1048 // ArabTeX, though, cannot handle this special behavior, it seems.
1049 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1050 || running_font.language()->lang() == "arabic_arabtex";
1051 if (open_font && !inset->inheritFont()) {
1052 bool needPar = false;
1053 bool closeLanguage = arabtex
1054 || basefont.isRightToLeft() == running_font.isRightToLeft();
1055 unsigned int count = running_font.latexWriteEndChanges(os,
1056 bparams, runparams, basefont, basefont,
1057 needPar, closeLanguage);
1059 // if any font properties were closed, update the running_font,
1060 // making sure, however, to leave the language as it was
1062 // FIXME: probably a better way to keep track of the old
1063 // language, than copying the entire font?
1064 Font const copy_font(running_font);
1065 basefont = owner_->getLayoutFont(bparams, outerfont);
1066 running_font = basefont;
1068 running_font.setLanguage(copy_font.language());
1069 // leave font open if language is still open
1070 open_font = (running_font.language() == basefont.language());
1072 runparams.local_font = &basefont;
1076 size_t const previous_row_count = os.texrow().rows();
1079 runparams.lastid = id_;
1080 runparams.lastpos = i;
1081 inset->latex(os, runparams);
1082 } catch (EncodingException & e) {
1083 // add location information and throw again.
1092 if (os.texrow().rows() > previous_row_count) {
1093 os.texrow().start(owner_->id(), i + 1);
1096 column += (unsigned int)(os.os().tellp() - len);
1099 if (owner_->isDeleted(i))
1100 --runparams.inDeletedInset;
1104 void Paragraph::Private::latexSpecialChar(otexstream & os,
1105 BufferParams const & bparams,
1106 OutputParams const & runparams,
1107 Font const & running_font,
1108 string & alien_script,
1109 Layout const & style,
1112 unsigned int & column)
1114 char_type const c = owner_->getUChar(bparams, runparams, i);
1116 if (style.pass_thru || runparams.pass_thru
1117 || contains(style.pass_thru_chars, c)
1118 || contains(runparams.pass_thru_chars, c)) {
1120 Encoding const * const enc = runparams.encoding;
1121 if (enc && !enc->encodable(c))
1122 throw EncodingException(c);
1128 // TIPA uses its own T3 encoding
1129 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1131 // If T1 font encoding is used, use the special
1132 // characters it provides.
1133 // NOTE: Some languages reset the font encoding internally to a
1134 // non-standard font encoding. If we are using such a language,
1135 // we do not output special T1 chars.
1136 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1137 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1138 && latexSpecialT1(c, os, i, column))
1140 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1141 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1142 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1145 // Otherwise, we use what LaTeX provides us.
1148 os << "\\textbackslash" << termcmd;
1152 os << "\\textless" << termcmd;
1156 os << "\\textgreater" << termcmd;
1160 os << "\\textbar" << termcmd;
1165 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1166 (end_pos == -1 || i + 1 < end_pos) &&
1167 text_[i+1] == '-') {
1168 // Prevent "--" becoming an en dash and "---" an em dash.
1169 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1175 os << "\\textquotedbl" << termcmd;
1180 case '%': case '#': case '{':
1188 os << "\\textasciitilde" << termcmd;
1193 os << "\\textasciicircum" << termcmd;
1200 // avoid being mistaken for optional arguments
1208 // Blanks are printed before font switching.
1209 // Sure? I am not! (try nice-latex)
1210 // I am sure it's correct. LyX might be smarter
1211 // in the future, but for now, nothing wrong is
1217 // XeTeX's dash behaviour is determined via a global setting
1218 if (bparams.use_dash_ligatures
1219 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1221 // TODO #10961: && not in inset Flex Code
1222 // TODO #10961: && not in layout LyXCode
1223 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1240 Encoding const & encoding = *(runparams.encoding);
1241 char_type next = '\0';
1242 if (i + 1 < int(text_.size())) {
1243 next = text_[i + 1];
1244 if (Encodings::isCombiningChar(next)) {
1245 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1250 pair<docstring, bool> latex = encoding.latexChar(c);
1251 docstring nextlatex;
1252 bool nexttipas = false;
1253 string nexttipashortcut;
1254 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1255 nextlatex = encoding.latexChar(next).first;
1256 if (runparams.inIPA) {
1257 nexttipashortcut = Encodings::TIPAShortcut(next);
1258 nexttipas = !nexttipashortcut.empty();
1262 if (runparams.inIPA) {
1263 string const tipashortcut = Encodings::TIPAShortcut(c);
1264 if (!tipashortcut.empty()) {
1265 latex.first = from_ascii(tipashortcut);
1266 latex.second = false;
1270 // eventually close "script wrapper" command (see `Paragraph::latex`)
1271 if (!alien_script.empty()
1272 && alien_script != Encodings::isKnownScriptChar(next)) {
1273 column += latex.first.length();
1274 alien_script.clear();
1275 os << latex.first << "}";
1279 && ((!prefixIs(nextlatex, '\\')
1280 && !prefixIs(nextlatex, '{')
1281 && !prefixIs(nextlatex, '}'))
1283 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1285 // Prevent eating of a following space or command corruption by
1286 // following characters
1287 if (next == ' ' || next == '\0') {
1288 column += latex.first.length() + 1;
1289 os << latex.first << "{}";
1291 column += latex.first.length();
1292 os << latex.first << " ";
1295 column += latex.first.length() - 1;
1303 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1304 pos_type i, unsigned int & column)
1310 // In T1 encoding, these characters exist
1311 // but we should avoid ligatures
1312 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1314 os << "\\textcompwordmark" << termcmd;
1321 // soul.sty breaks with \char`\"
1322 os << "\\textquotedbl" << termcmd;
1331 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1332 pos_type i, unsigned int & column)
1334 // TU encoding is currently on par with T1.
1335 return latexSpecialT1(c, os, i, column);
1339 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1340 pos_type /*i*/, unsigned int & column)
1350 os << "\\textvertline" << termcmd;
1359 void Paragraph::Private::validate(LaTeXFeatures & features) const
1361 if (layout_->inpreamble && inset_owner_) {
1362 // FIXME: Using a string stream here circumvents the encoding
1363 // switching machinery of odocstream. Therefore the
1364 // output is wrong if this paragraph contains content
1365 // that needs to switch encoding.
1366 Buffer const & buf = inset_owner_->buffer();
1367 otexstringstream os;
1368 os << layout_->preamble();
1369 size_t const length = os.length();
1370 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1371 buf.getParFromID(owner_->id()).pit(), os,
1372 features.runparams(), string(), 0, -1, true);
1373 if (os.length() > length)
1374 features.addPreambleSnippet(os.release(), true);
1377 if (features.runparams().flavor == OutputParams::HTML
1378 && layout_->htmltitle()) {
1379 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1382 // check the params.
1383 if (!params_.spacing().isDefault())
1384 features.require("setspace");
1387 features.useLayout(layout_->name());
1390 fontlist_.validate(features);
1392 // then the indentation
1393 if (!params_.leftIndent().zero())
1394 features.require("ParagraphLeftIndent");
1397 InsetList::const_iterator icit = insetlist_.begin();
1398 InsetList::const_iterator iend = insetlist_.end();
1399 for (; icit != iend; ++icit) {
1401 features.inDeletedInset(owner_->isDeleted(icit->pos));
1402 if (icit->inset->lyxCode() == FOOT_CODE) {
1403 // FIXME: an item inset would make things much easier.
1404 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1405 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1406 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1407 && (icit->pos < begin_of_body_
1408 || (icit->pos == begin_of_body_
1409 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1410 features.saveNoteEnv("description");
1412 icit->inset->validate(features);
1413 features.inDeletedInset(false);
1414 if (layout_->needprotect &&
1415 icit->inset->lyxCode() == FOOT_CODE)
1416 features.require("NeedLyXFootnoteCode");
1420 // then the contents
1421 BufferParams const bp = features.runparams().is_child
1422 ? features.buffer().masterParams() : features.buffer().params();
1423 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1424 char_type c = text_[i];
1425 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1427 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1428 features.require("textquotedblp");
1429 else if (bp.main_font_encoding() != "T1"
1430 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1431 features.require("textquotedbl");
1432 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1433 // features that depend on the font or input encoding
1434 string feats = ci.textpreamble();
1435 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1436 if (fontenc.empty())
1437 fontenc = features.runparams().main_fontenc;
1438 while (!feats.empty()) {
1440 feats = split(feats, feat, ',');
1441 if (contains(feat, "!=")) {
1442 // a feature that is required except for the spcified
1443 // font or input encodings
1445 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1446 // multiple encodings are separated by semicolon
1447 vector<string> context = getVectorFromString(contexts, ";");
1448 // require feature if the context matches neither current font
1449 // nor input encoding
1450 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1451 && std::find(context.begin(), context.end(),
1452 features.runparams().encoding->name()) == context.end())
1453 features.require(realfeature);
1454 } else if (contains(feat, '=')) {
1455 // a feature that is required only for the spcified
1456 // font or input encodings
1458 string const contexts = split(feat, realfeature, '=');
1459 // multiple encodings are separated by semicolon
1460 vector<string> context = getVectorFromString(contexts, ";");
1461 // require feature if the context matches either current font
1462 // or input encoding
1463 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1464 || std::find(context.begin(), context.end(),
1465 features.runparams().encoding->name()) != context.end())
1466 features.require(realfeature);
1469 } else if (!bp.use_dash_ligatures
1470 && (c == 0x2013 || c == 0x2014)
1471 && bp.useNonTeXFonts
1472 && features.runparams().flavor == OutputParams::XETEX)
1473 // XeTeX's dash behaviour is determined via a global setting
1474 features.require("xetexdashbreakstate");
1475 BufferEncodings::validate(c, features);
1479 /////////////////////////////////////////////////////////////////////
1483 /////////////////////////////////////////////////////////////////////
1486 Layout const emptyParagraphLayout;
1489 Paragraph::Paragraph()
1490 : d(new Paragraph::Private(this, emptyParagraphLayout))
1497 Paragraph::Paragraph(Paragraph const & par)
1498 : itemdepth(par.itemdepth),
1499 d(new Paragraph::Private(*par.d, this))
1505 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1506 : itemdepth(par.itemdepth),
1507 d(new Paragraph::Private(*par.d, this, beg, end))
1513 Paragraph & Paragraph::operator=(Paragraph const & par)
1515 // needed as we will destroy the private part before copying it
1517 itemdepth = par.itemdepth;
1521 d = new Private(*par.d, this);
1528 Paragraph::~Paragraph()
1537 // this shall be called just before every "os << ..." action.
1538 void flushString(ostream & os, docstring & s)
1547 void Paragraph::write(ostream & os, BufferParams const & bparams,
1548 depth_type & dth) const
1550 // The beginning or end of a deeper (i.e. nested) area?
1551 if (dth != d->params_.depth()) {
1552 if (d->params_.depth() > dth) {
1553 while (d->params_.depth() > dth) {
1554 os << "\n\\begin_deeper";
1558 while (d->params_.depth() < dth) {
1559 os << "\n\\end_deeper";
1565 // First write the layout
1566 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1568 d->params_.write(os);
1570 Font font1(inherit_font, bparams.language);
1572 Change running_change = Change(Change::UNCHANGED);
1574 // this string is used as a buffer to avoid repetitive calls
1575 // to to_utf8(), which turn out to be expensive (JMarc)
1576 docstring write_buffer;
1579 for (pos_type i = 0; i <= size(); ++i) {
1581 Change const & change = lookupChange(i);
1582 if (change != running_change)
1583 flushString(os, write_buffer);
1584 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1585 running_change = change;
1590 // Write font changes
1591 Font font2 = getFontSettings(bparams, i);
1592 if (font2 != font1) {
1593 flushString(os, write_buffer);
1594 font2.lyxWriteChanges(font1, os);
1599 char_type const c = d->text_[i];
1602 if (Inset const * inset = getInset(i)) {
1603 flushString(os, write_buffer);
1604 if (inset->directWrite()) {
1605 // international char, let it write
1606 // code directly so it's shorter in
1612 os << "\\begin_inset ";
1614 os << "\n\\end_inset\n\n";
1617 // FIXME This can be removed again once the mystery
1618 // crash has been resolved.
1623 flushString(os, write_buffer);
1624 os << "\n\\backslash\n";
1628 flushString(os, write_buffer);
1629 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1636 if ((column > 70 && c == ' ')
1638 flushString(os, write_buffer);
1642 // this check is to amend a bug. LyX sometimes
1643 // inserts '\0' this could cause problems.
1645 write_buffer.push_back(c);
1647 LYXERR0("NUL char in structure.");
1653 flushString(os, write_buffer);
1654 os << "\n\\end_layout\n";
1655 // FIXME This can be removed again once the mystery
1656 // crash has been resolved.
1661 void Paragraph::validate(LaTeXFeatures & features) const
1663 d->validate(features);
1664 bool fragile = features.runparams().moving_arg;
1665 fragile |= layout().needprotect;
1666 if (inInset().getLayout().isNeedProtect())
1668 if (needsCProtection(fragile))
1669 features.require("cprotect");
1673 void Paragraph::insert(pos_type start, docstring const & str,
1674 Font const & font, Change const & change)
1676 for (size_t i = 0, n = str.size(); i != n ; ++i)
1677 insertChar(start + i, str[i], font, change);
1681 void Paragraph::appendChar(char_type c, Font const & font,
1682 Change const & change)
1685 d->changes_.insert(change, d->text_.size());
1686 // when appending characters, no need to update tables
1687 d->text_.push_back(c);
1688 setFont(d->text_.size() - 1, font);
1689 d->requestSpellCheck(d->text_.size() - 1);
1693 void Paragraph::appendString(docstring const & s, Font const & font,
1694 Change const & change)
1696 pos_type end = s.size();
1697 size_t oldsize = d->text_.size();
1698 size_t newsize = oldsize + end;
1699 size_t capacity = d->text_.capacity();
1700 if (newsize >= capacity)
1701 d->text_.reserve(max(capacity + 100, newsize));
1703 // when appending characters, no need to update tables
1706 // FIXME: Optimize this!
1707 for (size_t i = oldsize; i != newsize; ++i) {
1709 d->changes_.insert(change, i);
1710 d->requestSpellCheck(i);
1712 d->fontlist_.set(oldsize, font);
1713 d->fontlist_.set(newsize - 1, font);
1717 void Paragraph::insertChar(pos_type pos, char_type c,
1720 d->insertChar(pos, c, Change(trackChanges ?
1721 Change::INSERTED : Change::UNCHANGED));
1725 void Paragraph::insertChar(pos_type pos, char_type c,
1726 Font const & font, bool trackChanges)
1728 d->insertChar(pos, c, Change(trackChanges ?
1729 Change::INSERTED : Change::UNCHANGED));
1734 void Paragraph::insertChar(pos_type pos, char_type c,
1735 Font const & font, Change const & change)
1737 d->insertChar(pos, c, change);
1742 void Paragraph::resetFonts(Font const & font)
1744 d->fontlist_.clear();
1745 d->fontlist_.set(0, font);
1746 d->fontlist_.set(d->text_.size() - 1, font);
1749 // Gets uninstantiated font setting at position.
1750 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1754 LYXERR0("pos: " << pos << " size: " << size());
1758 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1759 if (cit != d->fontlist_.end())
1762 if (pos == size() && !empty())
1763 return getFontSettings(bparams, pos - 1);
1765 // Optimisation: avoid a full font instantiation if there is no
1766 // language change from previous call.
1767 static Font previous_font;
1768 static Language const * previous_lang = nullptr;
1769 Language const * lang = getParLanguage(bparams);
1770 if (lang != previous_lang) {
1771 previous_lang = lang;
1772 previous_font = Font(inherit_font, lang);
1774 return previous_font;
1778 FontSpan Paragraph::fontSpan(pos_type pos) const
1780 LBUFERR(pos <= size());
1783 return FontSpan(pos, pos);
1786 FontList::const_iterator cit = d->fontlist_.begin();
1787 FontList::const_iterator end = d->fontlist_.end();
1788 for (; cit != end; ++cit) {
1789 if (cit->pos() >= pos) {
1790 if (pos >= beginOfBody())
1791 return FontSpan(max(start, beginOfBody()),
1794 return FontSpan(start,
1795 min(beginOfBody() - 1,
1798 start = cit->pos() + 1;
1801 // This should not happen, but if so, we take no chances.
1802 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1803 LASSERT(false, return FontSpan(pos, pos));
1807 // Gets uninstantiated font setting at position 0
1808 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1810 if (!empty() && !d->fontlist_.empty())
1811 return d->fontlist_.begin()->font();
1813 // Optimisation: avoid a full font instantiation if there is no
1814 // language change from previous call.
1815 static Font previous_font;
1816 static Language const * previous_lang = nullptr;
1817 if (bparams.language != previous_lang) {
1818 previous_lang = bparams.language;
1819 previous_font = Font(inherit_font, bparams.language);
1822 return previous_font;
1826 // Gets the fully instantiated font at a given position in a paragraph
1827 // This is basically the same function as Text::GetFont() in text2.cpp.
1828 // The difference is that this one is used for generating the LaTeX file,
1829 // and thus cosmetic "improvements" are disallowed: This has to deliver
1830 // the true picture of the buffer. (Asger)
1831 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1832 Font const & outerfont) const
1836 Font font = getFontSettings(bparams, pos);
1838 pos_type const body_pos = beginOfBody();
1839 FontInfo & fi = font.fontInfo();
1841 fi.realize(d->layout_->labelfont);
1843 fi.realize(d->layout_->font);
1845 fi.realize(outerfont.fontInfo());
1846 fi.realize(bparams.getFont().fontInfo());
1852 Font const Paragraph::getLabelFont
1853 (BufferParams const & bparams, Font const & outerfont) const
1855 FontInfo tmpfont = d->layout_->labelfont;
1856 tmpfont.realize(outerfont.fontInfo());
1857 tmpfont.realize(bparams.getFont().fontInfo());
1858 return Font(tmpfont, getParLanguage(bparams));
1862 Font const Paragraph::getLayoutFont
1863 (BufferParams const & bparams, Font const & outerfont) const
1865 FontInfo tmpfont = d->layout_->font;
1866 tmpfont.realize(outerfont.fontInfo());
1867 tmpfont.realize(bparams.getFont().fontInfo());
1868 return Font(tmpfont, getParLanguage(bparams));
1872 char_type Paragraph::getUChar(BufferParams const & bparams,
1873 OutputParams const & rp,
1876 char_type c = d->text_[pos];
1878 // Return unchanged character in LTR languages
1879 // or if we use poylglossia/bidi (XeTeX).
1880 if (rp.useBidiPackage()
1881 || !getFontSettings(bparams, pos).isRightToLeft())
1884 // Without polyglossia/bidi, we need to account for some special cases.
1885 // FIXME This needs to be audited!
1887 // * The input is as expected for all delimiters
1888 // => checked for Hebrew!
1889 // * The output matches the display in the LyX workarea
1890 // => checked for Hebrew!
1891 // * The special cases below are really necessary
1892 // => checked for Hebrew!
1893 // * In arabic_arabi, brackets are transformed to Arabic
1894 // Ornate Parentheses. Is this is really wanted?
1896 string const & lang = getFontSettings(bparams, pos).language()->lang();
1899 // 1. In the following languages, parentheses need to be reversed.
1900 // Also with polyglodia/luabidi
1901 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1903 // 2. In the following languages, brackets don't need to be reversed.
1904 bool const reversebrackets = lang != "arabic_arabtex"
1905 && lang != "arabic_arabi"
1908 // Now swap delimiters if needed.
1919 if (reversebrackets)
1923 if (reversebrackets)
1944 void Paragraph::setFont(pos_type pos, Font const & font)
1946 LASSERT(pos <= size(), return);
1948 // First, reduce font against layout/label font
1949 // Update: The setCharFont() routine in text2.cpp already
1950 // reduces font, so we don't need to do that here. (Asger)
1952 d->fontlist_.set(pos, font);
1956 void Paragraph::makeSameLayout(Paragraph const & par)
1958 d->layout_ = par.d->layout_;
1959 d->params_ = par.d->params_;
1963 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1965 if (isFreeSpacing())
1971 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1972 if (eraseChar(pos, trackChanges))
1978 return count > 0 || pos > 0;
1982 bool Paragraph::hasSameLayout(Paragraph const & par) const
1984 return par.d->layout_ == d->layout_
1985 && d->params_.sameLayout(par.d->params_);
1989 depth_type Paragraph::getDepth() const
1991 return d->params_.depth();
1995 depth_type Paragraph::getMaxDepthAfter() const
1997 if (d->layout_->isEnvironment())
1998 return d->params_.depth() + 1;
2000 return d->params_.depth();
2004 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2006 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2007 return getDefaultAlign(bparams);
2009 return d->params_.align();
2013 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2015 LyXAlignment res = layout().align;
2016 if (isRTL(bparams)) {
2018 if (res == LYX_ALIGN_LEFT)
2019 res = LYX_ALIGN_RIGHT;
2020 else if (res == LYX_ALIGN_RIGHT)
2021 res = LYX_ALIGN_LEFT;
2027 docstring const & Paragraph::labelString() const
2029 return d->params_.labelString();
2033 // the next two functions are for the manual labels
2034 docstring const Paragraph::getLabelWidthString() const
2036 if (d->layout_->margintype == MARGIN_MANUAL
2037 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2038 return d->params_.labelWidthString();
2040 return _("Senseless with this layout!");
2044 void Paragraph::setLabelWidthString(docstring const & s)
2046 d->params_.labelWidthString(s);
2050 docstring Paragraph::expandLabel(Layout const & layout,
2051 BufferParams const & bparams) const
2053 return expandParagraphLabel(layout, bparams, true);
2057 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2058 BufferParams const & bparams) const
2060 return expandParagraphLabel(layout, bparams, false);
2064 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2065 BufferParams const & bparams, bool process_appendix) const
2067 DocumentClass const & tclass = bparams.documentClass();
2068 string const & lang = getParLanguage(bparams)->code();
2069 bool const in_appendix = process_appendix && d->params_.appendix();
2070 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2072 if (fmt.empty() && !layout.counter.empty())
2073 return tclass.counters().theCounter(layout.counter, lang);
2075 // handle 'inherited level parts' in 'fmt',
2076 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2077 size_t const i = fmt.find('@', 0);
2078 if (i != docstring::npos) {
2079 size_t const j = fmt.find('@', i + 1);
2080 if (j != docstring::npos) {
2081 docstring parent(fmt, i + 1, j - i - 1);
2082 docstring label = from_ascii("??");
2083 if (tclass.hasLayout(parent))
2084 label = expandParagraphLabel(tclass[parent], bparams,
2086 fmt = docstring(fmt, 0, i) + label
2087 + docstring(fmt, j + 1, docstring::npos);
2091 return tclass.counters().counterLabel(fmt, lang);
2095 void Paragraph::applyLayout(Layout const & new_layout)
2097 d->layout_ = &new_layout;
2098 LyXAlignment const oldAlign = d->params_.align();
2100 if (!(oldAlign & d->layout_->alignpossible)) {
2101 frontend::Alert::warning(_("Alignment not permitted"),
2102 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2103 d->params_.align(LYX_ALIGN_LAYOUT);
2108 pos_type Paragraph::beginOfBody() const
2110 return d->begin_of_body_;
2114 void Paragraph::setBeginOfBody()
2116 if (d->layout_->labeltype != LABEL_MANUAL) {
2117 d->begin_of_body_ = 0;
2121 // Unroll the first two cycles of the loop
2122 // and remember the previous character to
2123 // remove unnecessary getChar() calls
2125 pos_type end = size();
2126 bool prev_char_deleted = false;
2127 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2130 char_type previous_char = d->text_[i];
2131 if (!(isNewline(i) || isEnvSeparator(i))) {
2133 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2134 char_type temp = d->text_[i];
2135 prev_char_deleted = isDeleted(i);
2136 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2139 previous_char = temp;
2145 d->begin_of_body_ = i;
2149 bool Paragraph::allowParagraphCustomization() const
2151 return inInset().allowParagraphCustomization();
2155 bool Paragraph::usePlainLayout() const
2157 return inInset().usePlainLayout();
2161 bool Paragraph::isPassThru() const
2163 return inInset().isPassThru() || d->layout_->pass_thru;
2168 // paragraphs inside floats need different alignment tags to avoid
2171 bool noTrivlistCentering(InsetCode code)
2173 return code == FLOAT_CODE
2174 || code == WRAP_CODE
2175 || code == CELL_CODE;
2179 string correction(string const & orig)
2181 if (orig == "flushleft")
2182 return "raggedright";
2183 if (orig == "flushright")
2184 return "raggedleft";
2185 if (orig == "center")
2191 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2192 InsetCode code, bool const lastpar, int & col)
2194 string macro = suffix + "{";
2195 if (noTrivlistCentering(code)) {
2197 // the last paragraph in non-trivlist-aligned
2198 // context is special (to avoid unwanted whitespace)
2199 if (suffix == "\\begin") {
2200 macro = "\\" + correction(env) + "{}";
2201 os << from_ascii(macro);
2202 col += macro.size();
2207 macro += correction(env);
2211 if (suffix == "\\par\\end") {
2215 os << from_ascii(macro);
2216 col += macro.size();
2217 if (suffix == "\\begin") {
2227 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2228 otexstream & os, OutputParams const & runparams) const
2233 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2234 (layout_->toggle_indent != ITOGGLE_NEVER) :
2235 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2237 if (canindent && params_.noindent() && !layout_->pass_thru) {
2238 os << "\\noindent ";
2242 LyXAlignment const curAlign = params_.align();
2244 if (curAlign == layout_->align)
2248 case LYX_ALIGN_NONE:
2249 case LYX_ALIGN_BLOCK:
2250 case LYX_ALIGN_LAYOUT:
2251 case LYX_ALIGN_SPECIAL:
2252 case LYX_ALIGN_DECIMAL:
2254 case LYX_ALIGN_LEFT:
2255 case LYX_ALIGN_RIGHT:
2256 case LYX_ALIGN_CENTER:
2257 if (runparams.moving_arg) {
2264 string const begin_tag = "\\begin";
2265 InsetCode code = ownerCode();
2266 bool const lastpar = runparams.isLastPar;
2267 // RTL in classic (PDF)LaTeX (without the Bidi package)
2268 // Luabibdi (used by LuaTeX) behaves like classic
2269 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2270 && !runparams.useBidiPackage();
2273 case LYX_ALIGN_NONE:
2274 case LYX_ALIGN_BLOCK:
2275 case LYX_ALIGN_LAYOUT:
2276 case LYX_ALIGN_SPECIAL:
2277 case LYX_ALIGN_DECIMAL:
2279 case LYX_ALIGN_LEFT: {
2281 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2282 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2284 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2286 } case LYX_ALIGN_RIGHT: {
2288 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2289 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2291 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2293 } case LYX_ALIGN_CENTER: {
2294 corrected_env(os, begin_tag, "center", code, lastpar, column);
2303 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2304 otexstream & os, OutputParams const & runparams) const
2306 LyXAlignment const curAlign = params_.align();
2308 if (curAlign == layout_->align)
2312 case LYX_ALIGN_NONE:
2313 case LYX_ALIGN_BLOCK:
2314 case LYX_ALIGN_LAYOUT:
2315 case LYX_ALIGN_SPECIAL:
2316 case LYX_ALIGN_DECIMAL:
2318 case LYX_ALIGN_LEFT:
2319 case LYX_ALIGN_RIGHT:
2320 case LYX_ALIGN_CENTER:
2321 if (runparams.moving_arg)
2326 bool output = false;
2328 string const end_tag = "\\par\\end";
2329 InsetCode code = ownerCode();
2330 bool const lastpar = runparams.isLastPar;
2331 // RTL in classic (PDF)LaTeX (without the Bidi package)
2332 // Luabibdi (used by LuaTeX) behaves like classic
2333 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2334 && !runparams.useBidiPackage();
2337 case LYX_ALIGN_NONE:
2338 case LYX_ALIGN_BLOCK:
2339 case LYX_ALIGN_LAYOUT:
2340 case LYX_ALIGN_SPECIAL:
2341 case LYX_ALIGN_DECIMAL:
2343 case LYX_ALIGN_LEFT: {
2345 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2346 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2348 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2350 } case LYX_ALIGN_RIGHT: {
2352 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2353 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2355 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2357 } case LYX_ALIGN_CENTER: {
2358 corrected_env(os, end_tag, "center", code, lastpar, col);
2363 return output || lastpar;
2367 // This one spits out the text of the paragraph
2368 void Paragraph::latex(BufferParams const & bparams,
2369 Font const & outerfont,
2371 OutputParams const & runparams,
2372 int start_pos, int end_pos, bool force) const
2374 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2376 // FIXME This check should not be needed. Perhaps issue an
2377 // error if it triggers.
2378 Layout const & style = inInset().forcePlainLayout() ?
2379 bparams.documentClass().plainLayout() : *d->layout_;
2381 if (!force && style.inpreamble)
2384 bool const allowcust = allowParagraphCustomization();
2386 // Current base font for all inherited font changes, without any
2387 // change caused by an individual character, except for the language:
2388 // It is set to the language of the first character.
2389 // As long as we are in the label, this font is the base font of the
2390 // label. Before the first body character it is set to the base font
2394 // If there is an open font-encoding changing command (script wrapper),
2395 // alien_script is set to its name
2396 string alien_script;
2399 // Maybe we have to create a optional argument.
2400 pos_type body_pos = beginOfBody();
2401 unsigned int column = 0;
2404 // the optional argument is kept in curly brackets in
2405 // case it contains a ']'
2406 // This is not strictly needed, but if this is changed it
2407 // would be a file format change, and tex2lyx would need
2408 // to be adjusted, since it unconditionally removes the
2409 // braces when it parses \item.
2412 basefont = getLabelFont(bparams, outerfont);
2414 basefont = getLayoutFont(bparams, outerfont);
2417 // Which font is currently active?
2418 Font running_font(basefont);
2419 // Do we have an open font change?
2420 bool open_font = false;
2422 Change runningChange = Change(Change::UNCHANGED);
2424 Encoding const * const prev_encoding = runparams.encoding;
2426 os.texrow().start(id(), 0);
2428 // if the paragraph is empty, the loop will not be entered at all
2430 // For InTitle commands, we have already opened a group
2431 // in output_latex::TeXOnePar.
2432 if (style.isCommand() && !style.intitle) {
2436 if (!style.leftdelim().empty()) {
2437 os << style.leftdelim();
2438 column += style.leftdelim().size();
2441 column += d->startTeXParParams(bparams, os, runparams);
2444 // Whether a \par can be issued for insets typeset inline with text.
2445 // Yes if greater than 0. This has to be static.
2446 THREAD_LOCAL_STATIC int parInline = 0;
2448 for (pos_type i = 0; i < size(); ++i) {
2449 // First char in paragraph or after label?
2450 if (i == body_pos) {
2453 bool needPar = false;
2454 column += running_font.latexWriteEndChanges(
2455 os, bparams, runparams,
2456 basefont, basefont, needPar);
2459 basefont = getLayoutFont(bparams, outerfont);
2460 running_font = basefont;
2462 column += Changes::latexMarkChange(os, bparams,
2463 runningChange, Change(Change::UNCHANGED),
2465 runningChange = Change(Change::UNCHANGED);
2470 // For InTitle commands, we have already opened a group
2471 // in output_latex::TeXOnePar.
2472 if (style.isCommand() && !style.intitle) {
2477 if (!style.leftdelim().empty()) {
2478 os << style.leftdelim();
2479 column += style.leftdelim().size();
2483 column += d->startTeXParParams(bparams, os,
2487 runparams.wasDisplayMath = runparams.inDisplayMath;
2488 runparams.inDisplayMath = false;
2489 bool deleted_display_math = false;
2490 Change const & change = runparams.inDeletedInset
2491 ? runparams.changeOfDeletedInset : lookupChange(i);
2493 char_type const c = d->text_[i];
2495 // Check whether a display math inset follows
2497 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2499 runparams.ctObject = getInset(i)->CtObject(runparams);
2501 InsetMath const * im = getInset(i)->asInsetMath();
2502 if (im && im->asHullInset()
2503 && im->asHullInset()->outerDisplay()) {
2504 runparams.inDisplayMath = true;
2505 // runparams.inDeletedInset will be set by
2506 // latexInset later, but we need this info
2507 // before it is called. On the other hand, we
2508 // cannot set it here because it is a counter.
2509 deleted_display_math = isDeleted(i);
2511 if (bparams.output_changes && deleted_display_math
2512 && runningChange == change
2513 && change.type == Change::DELETED
2514 && !os.afterParbreak()) {
2515 // A display math in the same paragraph follows.
2516 // We have to close and then reopen \lyxdeleted,
2517 // otherwise the math will be shifted up.
2518 OutputParams rp = runparams;
2520 bool needPar = false;
2521 column += running_font.latexWriteEndChanges(
2522 os, bparams, rp, basefont,
2526 basefont = (body_pos > i) ? getLabelFont(bparams, outerfont)
2527 : getLayoutFont(bparams, outerfont);
2528 running_font = basefont;
2529 column += Changes::latexMarkChange(os, bparams,
2530 Change(Change::INSERTED), change, rp);
2534 if (bparams.output_changes && runningChange != change) {
2535 if (!alien_script.empty()) {
2538 alien_script.clear();
2541 bool needPar = false;
2542 column += running_font.latexWriteEndChanges(
2543 os, bparams, runparams,
2544 basefont, basefont, needPar);
2547 basefont = (body_pos > i) ? getLabelFont(bparams, outerfont)
2548 : getLayoutFont(bparams, outerfont);
2549 running_font = basefont;
2550 column += Changes::latexMarkChange(os, bparams, runningChange,
2552 runningChange = change;
2555 // do not output text which is marked deleted
2556 // if change tracking output is disabled
2557 if (!bparams.output_changes && change.deleted()) {
2563 // Fully instantiated font
2564 Font const current_font = getFont(bparams, i, outerfont);
2566 Font const last_font = running_font;
2567 bool const in_ct_deletion = (bparams.output_changes
2568 && runningChange == change
2569 && change.type == Change::DELETED
2570 && !os.afterParbreak());
2572 // Do we need to close the previous font?
2574 (current_font != running_font ||
2575 current_font.language() != running_font.language()))
2577 // ensure there is no open script-wrapper
2578 if (!alien_script.empty()) {
2581 alien_script.clear();
2583 bool needPar = false;
2584 if (in_ct_deletion) {
2585 // We have to close and then reopen \lyxdeleted,
2586 // as strikeout needs to be on lowest level.
2590 column += running_font.latexWriteEndChanges(
2591 os, bparams, runparams, basefont,
2592 (i == body_pos-1) ? basefont : current_font,
2594 if (in_ct_deletion) {
2595 // We have to close and then reopen \lyxdeleted,
2596 // as strikeout needs to be on lowest level.
2597 OutputParams rp = runparams;
2598 column += Changes::latexMarkChange(os, bparams,
2599 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2601 running_font = basefont;
2605 // if necessary, close language environment before opening CJK
2606 string const running_lang = running_font.language()->babel();
2607 string const lang_end_command = lyxrc.language_command_end;
2608 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2609 && !running_lang.empty()
2610 && running_lang == openLanguageName()
2611 && current_font.language()->encoding()->package() == Encoding::CJK) {
2612 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2613 os << from_ascii(end_tag);
2614 column += end_tag.length();
2618 // Switch file encoding if necessary (and allowed)
2619 if (!runparams.pass_thru && !style.pass_thru &&
2620 runparams.encoding->package() != Encoding::none &&
2621 current_font.language()->encoding()->package() != Encoding::none) {
2622 pair<bool, int> const enc_switch =
2623 switchEncoding(os.os(), bparams, runparams,
2624 *(current_font.language()->encoding()));
2625 if (enc_switch.first) {
2626 column += enc_switch.second;
2627 runparams.encoding = current_font.language()->encoding();
2631 // A display math inset inside an ulem command will be output
2632 // as a box of width \linewidth, so we have to either disable
2633 // indentation if the inset starts a paragraph, or start a new
2634 // line to accommodate such box. This has to be done before
2635 // writing any font changing commands.
2636 if (runparams.inDisplayMath && !deleted_display_math
2637 && runparams.inulemcmd) {
2638 if (os.afterParbreak())
2644 // Do we need to change font?
2645 if ((current_font != running_font ||
2646 current_font.language() != running_font.language())
2647 && i != body_pos - 1)
2649 if (in_ct_deletion) {
2650 // We have to close and then reopen \lyxdeleted,
2651 // as strikeout needs to be on lowest level.
2652 bool needPar = false;
2653 OutputParams rp = runparams;
2654 column += running_font.latexWriteEndChanges(
2655 os, bparams, rp, basefont,
2660 odocstringstream ods;
2661 column += current_font.latexWriteStartChanges(ods, bparams,
2662 runparams, basefont,
2664 // Check again for display math in ulem commands as a
2665 // font change may also occur just before a math inset.
2666 if (runparams.inDisplayMath && !deleted_display_math
2667 && runparams.inulemcmd) {
2668 if (os.afterParbreak())
2673 running_font = current_font;
2675 docstring fontchange = ods.str();
2676 // check whether the fontchange ends with a \\textcolor
2677 // modifier and the text starts with a space (bug 4473)
2678 docstring const last_modifier = rsplit(fontchange, '\\');
2679 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2680 os << fontchange << from_ascii("{}");
2681 // check if the fontchange ends with a trailing blank
2682 // (like "\small " (see bug 3382)
2683 else if (suffixIs(fontchange, ' ') && c == ' ')
2684 os << fontchange.substr(0, fontchange.size() - 1)
2685 << from_ascii("{}");
2688 if (in_ct_deletion) {
2689 // We have to close and then reopen \lyxdeleted,
2690 // as strikeout needs to be on lowest level.
2691 OutputParams rp = runparams;
2692 column += Changes::latexMarkChange(os, bparams,
2693 Change(Change::UNCHANGED), change, rp);
2697 // FIXME: think about end_pos implementation...
2698 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2699 // FIXME: integrate this case in latexSpecialChar
2700 // Do not print the separation of the optional argument
2701 // if style.pass_thru is false. This works because
2702 // latexSpecialChar ignores spaces if
2703 // style.pass_thru is false.
2704 if (i != body_pos - 1) {
2705 if (d->simpleTeXBlanks(bparams, runparams, os,
2706 i, column, current_font, style)) {
2707 // A surrogate pair was output. We
2708 // must not call latexSpecialChar
2709 // in this iteration, since it would output
2710 // the combining character again.
2717 OutputParams rp = runparams;
2718 rp.free_spacing = style.free_spacing;
2719 rp.local_font = ¤t_font;
2720 rp.intitle = style.intitle;
2722 // Two major modes: LaTeX or plain
2723 // Handle here those cases common to both modes
2724 // and then split to handle the two modes separately.
2725 if (c == META_INSET) {
2726 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2727 // Greyedout notes and, in general, all insets
2728 // with InsetLayout::isDisplay() == false,
2729 // are typeset inline with the text. So, we
2730 // can add a \par to the last paragraph of
2731 // such insets only if nothing else follows.
2732 bool incremented = false;
2733 Inset const * inset = getInset(i);
2734 InsetText const * textinset = inset
2735 ? inset->asInsetText()
2737 if (i + 1 == size() && textinset
2738 && !inset->getLayout().isDisplay()) {
2739 ParagraphList const & pars =
2740 textinset->text().paragraphs();
2741 pit_type const pit = pars.size() - 1;
2742 Font const lastfont =
2743 pit < 0 || pars[pit].empty()
2744 ? pars[pit].getLayoutFont(
2747 : pars[pit].getFont(bparams,
2748 pars[pit].size() - 1,
2750 if (lastfont.fontInfo().size() !=
2751 basefont.fontInfo().size()) {
2756 d->latexInset(bparams, os, rp, running_font,
2757 basefont, outerfont, open_font,
2758 runningChange, style, i, column);
2762 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2763 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2764 // Close \lyx*deleted and force its
2765 // reopening (if needed)
2768 runningChange = Change(Change::UNCHANGED);
2769 runparams.ctObject = OutputParams::CT_NORMAL;
2772 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2773 if (!bparams.useNonTeXFonts)
2774 script = Encodings::isKnownScriptChar(c);
2775 if (script != alien_script) {
2776 if (!alien_script.empty()) {
2778 alien_script.clear();
2780 string fontenc = running_font.language()->fontenc(bparams);
2782 && !Encodings::fontencSupportsScript(fontenc, script)) {
2783 column += script.length() + 2;
2784 os << "\\" << script << "{";
2785 alien_script = script;
2789 d->latexSpecialChar(os, bparams, rp, running_font,
2790 alien_script, style, i, end_pos, column);
2791 } catch (EncodingException & e) {
2792 if (runparams.dryrun) {
2793 os << "<" << _("LyX Warning: ")
2794 << _("uncodable character") << " '";
2798 // add location information and throw again.
2806 // Set the encoding to that returned from latexSpecialChar (see
2807 // comment for encoding member in OutputParams.h)
2808 runparams.encoding = rp.encoding;
2810 // Also carry on the info on a closed ulem command for insets
2811 // such as Note that do not produce any output, so that no
2812 // command is ever executed but its opening was recorded.
2813 runparams.inulemcmd = rp.inulemcmd;
2815 // And finally, pass the post_macros upstream
2816 runparams.post_macro = rp.post_macro;
2819 // Close wrapper for alien script
2820 if (!alien_script.empty()) {
2822 alien_script.clear();
2825 // If we have an open font definition, we have to close it
2827 // Make sure that \\par is done with the font of the last
2828 // character if this has another size as the default.
2829 // This is necessary because LaTeX (and LyX on the screen)
2830 // calculates the space between the baselines according
2831 // to this font. (Matthias)
2833 // We must not change the font for the last paragraph
2834 // of non-multipar insets, tabular cells or commands,
2835 // since this produces unwanted whitespace.
2837 Font const font = empty()
2838 ? getLayoutFont(bparams, outerfont)
2839 : getFont(bparams, size() - 1, outerfont);
2841 InsetText const * textinset = inInset().asInsetText();
2843 bool const maintext = textinset
2844 ? textinset->text().isMainText()
2847 size_t const numpars = textinset
2848 ? textinset->text().paragraphs().size()
2851 bool needPar = false;
2853 if (style.resfont.size() != font.fontInfo().size()
2854 && (!runparams.isLastPar || maintext
2855 || (numpars > 1 && d->ownerCode() != CELL_CODE
2856 && (inInset().getLayout().isDisplay()
2858 && !style.isCommand()) {
2861 #ifdef FIXED_LANGUAGE_END_DETECTION
2863 running_font.latexWriteEndChanges(os, bparams,
2864 runparams, basefont,
2865 next_->getFont(bparams, 0, outerfont),
2868 running_font.latexWriteEndChanges(os, bparams,
2869 runparams, basefont, basefont, needPar);
2872 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2873 //FIXME: there as we start another \selectlanguage with the next paragraph if
2874 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2875 running_font.latexWriteEndChanges(os, bparams, runparams,
2876 basefont, basefont, needPar);
2879 // The \par could not be inserted at the same nesting
2880 // level of the font size change, so do it now.
2881 os << "{\\" << font.latexSize() << "\\par}";
2885 column += Changes::latexMarkChange(os, bparams, runningChange,
2886 Change(Change::UNCHANGED), runparams);
2888 // Needed if there is an optional argument but no contents.
2889 if (body_pos > 0 && body_pos == size()) {
2893 if (!style.rightdelim().empty()) {
2894 os << style.rightdelim();
2895 column += style.rightdelim().size();
2898 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2899 && runparams.encoding != prev_encoding) {
2900 runparams.encoding = prev_encoding;
2901 os << setEncoding(prev_encoding->iconvName());
2904 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2908 bool Paragraph::emptyTag() const
2910 for (pos_type i = 0; i < size(); ++i) {
2911 if (Inset const * inset = getInset(i)) {
2912 InsetCode lyx_code = inset->lyxCode();
2913 // FIXME testing like that is wrong. What is
2915 if (lyx_code != TOC_CODE &&
2916 lyx_code != INCLUDE_CODE &&
2917 lyx_code != GRAPHICS_CODE &&
2918 lyx_code != ERT_CODE &&
2919 lyx_code != LISTINGS_CODE &&
2920 lyx_code != FLOAT_CODE &&
2921 lyx_code != TABULAR_CODE) {
2925 char_type c = d->text_[i];
2926 if (c != ' ' && c != '\t')
2934 string Paragraph::getID(Buffer const &, OutputParams const &)
2937 for (pos_type i = 0; i < size(); ++i) {
2938 if (Inset const * inset = getInset(i)) {
2939 InsetCode lyx_code = inset->lyxCode();
2940 if (lyx_code == LABEL_CODE) {
2941 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2942 docstring const & id = il->getParam("name");
2943 return "id='" + to_utf8(xml::cleanID(id)) + "'";
2951 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2955 for (i = 0; i < size(); ++i) {
2956 if (Inset const * inset = getInset(i)) {
2957 inset->docbook(os, runparams);
2959 char_type c = d->text_[i];
2962 os << xml::escapeChar(c, XMLStream::ESCAPE_ALL);
2969 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
2973 for (i = 0; i < size(); ++i) {
2974 if (Inset const * inset = getInset(i)) {
2975 inset->xhtml(xs, runparams);
2977 char_type c = d->text_[i];
2987 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2990 pos_type size = text_.size();
2991 for (pos_type i = initial; i < size; ++i) {
2992 Font font = owner_->getFont(buf.params(), i, outerfont);
2993 if (text_[i] == META_INSET)
2995 if (i != initial && font != font_old)
3004 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3006 OutputParams const & runparams,
3007 Font const & outerfont,
3008 pos_type initial) const
3010 bool emph_flag = false;
3012 Layout const & style = *d->layout_;
3014 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3016 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3019 // parsing main loop
3020 for (pos_type i = initial; i < size(); ++i) {
3021 Font font = getFont(buf.params(), i, outerfont);
3023 // handle <emphasis> tag
3024 if (font_old.emph() != font.fontInfo().emph()) {
3025 if (font.fontInfo().emph() == FONT_ON) {
3028 } else if (i != initial) {
3029 os << "</emphasis>";
3034 if (Inset const * inset = getInset(i)) {
3035 inset->docbook(os, runparams);
3037 char_type c = d->text_[i];
3039 if (style.pass_thru)
3042 os << xml::escapeChar(c, XMLStream::EscapeSettings::ESCAPE_ALL);
3044 font_old = font.fontInfo();
3048 os << "</emphasis>";
3051 if (style.free_spacing)
3053 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3060 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3061 vector<xml::EndFontTag> & tagsToClose,
3062 bool & flag, FontState curstate, xml::FontTypes type)
3064 if (curstate == FONT_ON) {
3065 tagsToOpen.push_back(xhtmlStartFontTag(type));
3068 tagsToClose.push_back(xhtmlEndFontTag(type));
3073 } // anonymous namespace
3076 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3078 OutputParams const & runparams,
3079 Font const & outerfont,
3080 bool start_paragraph, bool close_paragraph,
3081 pos_type initial) const
3085 // track whether we have opened these tags
3086 bool emph_flag = false;
3087 bool bold_flag = false;
3088 bool noun_flag = false;
3089 bool ubar_flag = false;
3090 bool dbar_flag = false;
3091 bool sout_flag = false;
3092 bool xout_flag = false;
3093 bool wave_flag = false;
3095 bool shap_flag = false;
3097 bool faml_flag = false;
3099 bool size_flag = false;
3101 Layout const & style = *d->layout_;
3103 if (start_paragraph)
3104 xs.startDivision(allowEmpty());
3107 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3109 FontShape curr_fs = INHERIT_SHAPE;
3110 FontFamily curr_fam = INHERIT_FAMILY;
3111 FontSize curr_size = INHERIT_SIZE;
3113 string const default_family =
3114 buf.masterBuffer()->params().fonts_default_family;
3116 vector<xml::FontTag> tagsToOpen;
3117 vector<xml::EndFontTag> tagsToClose;
3119 // parsing main loop
3120 for (pos_type i = initial; i < size(); ++i) {
3121 // let's not show deleted material in the output
3125 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3128 FontState curstate = font.fontInfo().emph();
3129 if (font_old.emph() != curstate)
3130 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3133 curstate = font.fontInfo().noun();
3134 if (font_old.noun() != curstate)
3135 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3138 curstate = font.fontInfo().underbar();
3139 if (font_old.underbar() != curstate)
3140 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3143 curstate = font.fontInfo().strikeout();
3144 if (font_old.strikeout() != curstate)
3145 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3148 curstate = font.fontInfo().xout();
3149 if (font_old.xout() != curstate)
3150 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3153 curstate = font.fontInfo().uuline();
3154 if (font_old.uuline() != curstate)
3155 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3158 curstate = font.fontInfo().uwave();
3159 if (font_old.uwave() != curstate)
3160 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3163 // a little hackish, but allows us to reuse what we have.
3164 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3165 if (font_old.series() != font.fontInfo().series())
3166 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3169 curr_fs = font.fontInfo().shape();
3170 FontShape old_fs = font_old.shape();
3171 if (old_fs != curr_fs) {
3175 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_ITALIC), xml::FT_ITALIC));
3178 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_SLANTED), xml::FT_SLANTED));
3180 case SMALLCAPS_SHAPE:
3181 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_SMALLCAPS), xml::FT_SMALLCAPS));
3187 // the other tags are for internal use
3195 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_ITALIC), xml::FT_ITALIC));
3199 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_SLANTED), xml::FT_SLANTED));
3202 case SMALLCAPS_SHAPE:
3203 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_SMALLCAPS), xml::FT_SMALLCAPS));
3210 // the other tags are for internal use
3217 curr_fam = font.fontInfo().family();
3218 FontFamily old_fam = font_old.family();
3219 if (old_fam != curr_fam) {
3223 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_ROMAN), xml::FT_ROMAN));
3226 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_SANS), xml::FT_SANS));
3228 case TYPEWRITER_FAMILY:
3229 tagsToClose.push_back(xml::EndFontTag(fontToHtmlTag(xml::FT_TYPE), xml::FT_TYPE));
3231 case INHERIT_FAMILY:
3234 // the other tags are for internal use
3242 // we will treat a "default" font family as roman, since we have
3243 // no other idea what to do.
3244 if (default_family != "rmdefault" && default_family != "default") {
3245 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_ROMAN), xml::FT_ROMAN));
3250 if (default_family != "sfdefault") {
3251 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_SANS), xml::FT_SANS));
3255 case TYPEWRITER_FAMILY:
3256 if (default_family != "ttdefault") {
3257 tagsToOpen.push_back(xml::FontTag(fontToHtmlTag(xml::FT_TYPE), xml::FT_TYPE));
3261 case INHERIT_FAMILY:
3264 // the other tags are for internal use
3271 curr_size = font.fontInfo().size();
3272 FontSize old_size = font_old.size();
3273 if (old_size != curr_size) {
3277 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_TINY), xml::FT_SIZE_TINY);
3280 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_SCRIPT), xml::FT_SIZE_SCRIPT);
3283 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_FOOTNOTE), xml::FT_SIZE_FOOTNOTE);
3286 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_SMALL), xml::FT_SIZE_SMALL);
3289 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGE), xml::FT_SIZE_LARGE);
3292 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGER), xml::FT_SIZE_LARGER);
3295 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGEST), xml::FT_SIZE_LARGEST);
3298 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_HUGE), xml::FT_SIZE_HUGE);
3301 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_HUGER), xml::FT_SIZE_HUGER);
3304 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_INCREASE), xml::FT_SIZE_INCREASE);
3307 tagsToClose.emplace_back(fontToHtmlTag(xml::FT_SIZE_DECREASE), xml::FT_SIZE_DECREASE);
3313 // the other tags are for internal use
3319 switch (curr_size) {
3321 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_TINY), xml::FT_SIZE_TINY);
3325 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_SCRIPT), xml::FT_SIZE_SCRIPT);
3329 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_FOOTNOTE), xml::FT_SIZE_FOOTNOTE);
3333 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_SMALL), xml::FT_SIZE_SMALL);
3337 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGE), xml::FT_SIZE_LARGE);
3341 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGER), xml::FT_SIZE_LARGER);
3345 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_LARGEST), xml::FT_SIZE_LARGEST);
3349 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_HUGE), xml::FT_SIZE_HUGE);
3353 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_HUGER), xml::FT_SIZE_HUGER);
3357 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_INCREASE), xml::FT_SIZE_INCREASE);
3361 tagsToOpen.emplace_back(fontToHtmlTag(xml::FT_SIZE_DECREASE), xml::FT_SIZE_DECREASE);
3368 // the other tags are for internal use
3375 // Other such tags? What about the other text ranges?
3377 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3378 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3379 for (; cit != cen; ++cit)
3382 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3383 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3384 for (; sit != sen; ++sit)
3387 tagsToClose.clear();
3390 Inset const * inset = getInset(i);
3392 if (!runparams.for_toc || inset->isInToc()) {
3393 OutputParams np = runparams;
3394 np.local_font = &font;
3395 // If the paragraph has size 1, then we are in the "special
3396 // case" where we do not output the containing paragraph info
3397 if (!inset->getLayout().htmlisblock() && size() != 1)
3398 np.html_in_par = true;
3399 retval += inset->xhtml(xs, np);
3402 char_type c = getUChar(buf.masterBuffer()->params(),
3404 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3405 xs << XMLStream::ESCAPE_NONE << " ";
3409 font_old = font.fontInfo();
3413 // I'm worried about what happens if a branch, say, is itself
3414 // wrapped in some font stuff. I think that will not work.
3416 if (close_paragraph)
3423 bool Paragraph::isHfill(pos_type pos) const
3425 Inset const * inset = getInset(pos);
3426 return inset && inset->isHfill();
3430 bool Paragraph::isNewline(pos_type pos) const
3432 // U+2028 LINE SEPARATOR
3433 // U+2029 PARAGRAPH SEPARATOR
3434 char_type const c = d->text_[pos];
3435 if (c == 0x2028 || c == 0x2029)
3437 Inset const * inset = getInset(pos);
3438 return inset && inset->lyxCode() == NEWLINE_CODE;
3442 bool Paragraph::isEnvSeparator(pos_type pos) const
3444 Inset const * inset = getInset(pos);
3445 return inset && inset->lyxCode() == SEPARATOR_CODE;
3449 bool Paragraph::isLineSeparator(pos_type pos) const
3451 char_type const c = d->text_[pos];
3452 if (isLineSeparatorChar(c))
3454 Inset const * inset = getInset(pos);
3455 return inset && inset->isLineSeparator();
3459 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3463 if (ignore_deleted && isDeleted(pos))
3465 if (Inset const * inset = getInset(pos))
3466 return !inset->isLetter();
3467 // if we have a hard hyphen (no en- or emdash) or apostrophe
3468 // we pass this to the spell checker
3469 // FIXME: this method is subject to change, visit
3470 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3471 // to get an impression how complex this is.
3472 if (isHardHyphenOrApostrophe(pos))
3474 char_type const c = d->text_[pos];
3475 // We want to pass the escape chars to the spellchecker
3476 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3477 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3481 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3483 pos_type const psize = size();
3486 char_type const c = d->text_[pos];
3487 if (c != '-' && c != '\'')
3489 int nextpos = pos + 1;
3490 int prevpos = pos > 0 ? pos - 1 : 0;
3491 if ((nextpos == psize || isSpace(nextpos))
3492 && (pos == 0 || isSpace(prevpos)))
3498 bool Paragraph::needsCProtection(bool const fragile) const
3500 // first check the layout of the paragraph, but only in insets
3501 InsetText const * textinset = inInset().asInsetText();
3502 bool const maintext = textinset
3503 ? textinset->text().isMainText()
3506 if (!maintext && layout().needcprotect) {
3507 // Environments need cprotection regardless the content
3508 if (layout().latextype == LATEX_ENVIRONMENT)
3511 // Commands need cprotection if they contain specific chars
3512 int const nchars_escape = 9;
3513 static char_type const chars_escape[nchars_escape] = {
3514 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3516 docstring const pars = asString();
3517 for (int k = 0; k < nchars_escape; k++) {
3518 if (contains(pars, chars_escape[k]))
3523 // now check whether we have insets that need cprotection
3524 pos_type size = pos_type(d->text_.size());
3525 for (pos_type i = 0; i < size; ++i) {
3528 Inset const * ins = getInset(i);
3529 if (ins->needsCProtection(maintext, fragile))
3531 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3532 // Environments need cprotection regardless the content
3534 // Now check math environments
3535 InsetMath const * im = getInset(i)->asInsetMath();
3536 if (!im || im->cell(0).empty())
3538 switch(im->cell(0)[0]->lyxCode()) {
3539 case MATH_AMSARRAY_CODE:
3540 case MATH_SUBSTACK_CODE:
3542 case MATH_XYMATRIX_CODE:
3543 // these need cprotection
3554 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3556 return d->speller_state_.getRange(pos);
3560 bool Paragraph::isChar(pos_type pos) const
3562 if (Inset const * inset = getInset(pos))
3563 return inset->isChar();
3564 char_type const c = d->text_[pos];
3565 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3569 bool Paragraph::isSpace(pos_type pos) const
3571 if (Inset const * inset = getInset(pos))
3572 return inset->isSpace();
3573 char_type const c = d->text_[pos];
3574 return lyx::isSpace(c);
3579 Paragraph::getParLanguage(BufferParams const & bparams) const
3582 return getFirstFontSettings(bparams).language();
3583 // FIXME: we should check the prev par as well (Lgb)
3584 return bparams.language;
3588 bool Paragraph::isRTL(BufferParams const & bparams) const
3590 return getParLanguage(bparams)->rightToLeft()
3591 && !inInset().getLayout().forceLTR();
3595 void Paragraph::changeLanguage(BufferParams const & bparams,
3596 Language const * from, Language const * to)
3598 // change language including dummy font change at the end
3599 for (pos_type i = 0; i <= size(); ++i) {
3600 Font font = getFontSettings(bparams, i);
3601 if (font.language() == from) {
3602 font.setLanguage(to);
3604 d->requestSpellCheck(i);
3610 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3612 Language const * doc_language = bparams.language;
3613 for (auto const & f : d->fontlist_)
3614 if (f.font().language() != ignore_language &&
3615 f.font().language() != latex_language &&
3616 f.font().language() != doc_language)
3622 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3624 for (auto const & f : d->fontlist_) {
3625 Language const * lang = f.font().language();
3626 if (lang != ignore_language &&
3627 lang != latex_language)
3633 docstring Paragraph::asString(int options) const
3635 return asString(0, size(), options);
3639 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3641 odocstringstream os;
3644 && options & AS_STR_LABEL
3645 && !d->params_.labelString().empty())
3646 os << d->params_.labelString() << ' ';
3648 for (pos_type i = beg; i < end; ++i) {
3649 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3651 char_type const c = d->text_[i];
3652 if (isPrintable(c) || c == '\t'
3653 || (c == '\n' && (options & AS_STR_NEWLINES)))
3655 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3656 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3657 LASSERT(runparams != nullptr, return docstring());
3658 getInset(i)->plaintext(os, *runparams);
3660 getInset(i)->toString(os);
3669 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3670 bool const shorten, bool const label) const
3672 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3673 if (label && !labelString().empty())
3674 os += labelString() + ' ';
3675 if (!layout().isTocCaption())
3677 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3680 char_type const c = d->text_[i];
3683 else if (c == META_INSET)
3684 getInset(i)->forOutliner(os, tmplen, false);
3687 Text::shortenForOutliner(os, maxlen);
3691 void Paragraph::setInsetOwner(Inset const * inset)
3693 d->inset_owner_ = inset;
3697 int Paragraph::id() const
3703 void Paragraph::setId(int id)
3709 Layout const & Paragraph::layout() const
3715 void Paragraph::setLayout(Layout const & layout)
3717 d->layout_ = &layout;
3721 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3723 setLayout(tc.defaultLayout());
3727 void Paragraph::setPlainLayout(DocumentClass const & tc)
3729 setLayout(tc.plainLayout());
3733 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3735 if (usePlainLayout())
3736 setPlainLayout(tclass);
3738 setDefaultLayout(tclass);
3742 Inset const & Paragraph::inInset() const
3744 LBUFERR(d->inset_owner_);
3745 return *d->inset_owner_;
3749 ParagraphParameters & Paragraph::params()
3755 ParagraphParameters const & Paragraph::params() const
3761 bool Paragraph::isFreeSpacing() const
3763 if (d->layout_->free_spacing)
3765 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3769 bool Paragraph::allowEmpty() const
3771 if (d->layout_->keepempty)
3773 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3777 bool Paragraph::brokenBiblio() const
3779 // There is a problem if there is no bibitem at position 0 in
3780 // paragraphs that need one, if there is another bibitem in the
3781 // paragraph or if this paragraph is not supposed to have
3782 // a bibitem inset at all.
3783 return ((d->layout_->labeltype == LABEL_BIBLIO
3784 && (d->insetlist_.find(BIBITEM_CODE) != 0
3785 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
3786 || (d->layout_->labeltype != LABEL_BIBLIO
3787 && d->insetlist_.find(BIBITEM_CODE) != -1));
3791 int Paragraph::fixBiblio(Buffer const & buffer)
3793 // FIXME: when there was already an inset at 0, the return value is 1,
3794 // which does not tell whether another inset has been remove; the
3795 // cursor cannot be correctly updated.
3797 bool const track_changes = buffer.params().track_changes;
3798 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3800 // The case where paragraph is not BIBLIO
3801 if (d->layout_->labeltype != LABEL_BIBLIO) {
3802 if (bibitem_pos == -1)
3803 // No InsetBibitem => OK
3805 // There is an InsetBibitem: remove it!
3806 d->insetlist_.release(bibitem_pos);
3807 eraseChar(bibitem_pos, track_changes);
3808 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
3811 bool const hasbibitem0 = bibitem_pos == 0;
3813 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3814 // There was an InsetBibitem at pos 0,
3815 // and no other one => OK
3816 if (bibitem_pos == -1)
3818 // there is a bibitem at the 0 position, but since
3819 // there is a second one, we copy the second on the
3820 // first. We're assuming there are at most two of
3821 // these, which there should be.
3822 // FIXME: why does it make sense to do that rather
3823 // than keep the first? (JMarc)
3824 Inset * inset = releaseInset(bibitem_pos);
3825 d->insetlist_.begin()->inset = inset;
3826 return -bibitem_pos;
3829 // We need to create an inset at the beginning
3830 Inset * inset = nullptr;
3831 if (bibitem_pos > 0) {
3832 // there was one somewhere in the paragraph, let's move it
3833 inset = d->insetlist_.release(bibitem_pos);
3834 eraseChar(bibitem_pos, track_changes);
3837 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3838 InsetCommandParams(BIBITEM_CODE));
3840 Font font(inherit_font, buffer.params().language);
3841 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3842 : Change::UNCHANGED));
3844 // This is needed to get the counters right
3845 buffer.updateBuffer();
3850 void Paragraph::checkAuthors(AuthorList const & authorList)
3852 d->changes_.checkAuthors(authorList);
3856 bool Paragraph::isChanged(pos_type pos) const
3858 return lookupChange(pos).changed();
3862 bool Paragraph::isInserted(pos_type pos) const
3864 return lookupChange(pos).inserted();
3868 bool Paragraph::isDeleted(pos_type pos) const
3870 return lookupChange(pos).deleted();
3874 InsetList const & Paragraph::insetList() const
3876 return d->insetlist_;
3880 void Paragraph::setInsetBuffers(Buffer & b)
3882 d->insetlist_.setBuffer(b);
3886 void Paragraph::resetBuffer()
3888 d->insetlist_.resetBuffer();
3892 Inset * Paragraph::releaseInset(pos_type pos)
3894 Inset * inset = d->insetlist_.release(pos);
3895 /// does not honour change tracking!
3896 eraseChar(pos, false);
3901 Inset * Paragraph::getInset(pos_type pos)
3903 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3904 ? d->insetlist_.get(pos) : nullptr;
3908 Inset const * Paragraph::getInset(pos_type pos) const
3910 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3911 ? d->insetlist_.get(pos) : nullptr;
3915 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3916 pos_type & right, TextCase action)
3918 // process sequences of modified characters; in change
3919 // tracking mode, this approach results in much better
3920 // usability than changing case on a char-by-char basis
3921 // We also need to track the current font, since font
3922 // changes within sequences can occur.
3923 vector<pair<char_type, Font> > changes;
3925 bool const trackChanges = bparams.track_changes;
3927 bool capitalize = true;
3929 for (; pos < right; ++pos) {
3930 char_type oldChar = d->text_[pos];
3931 char_type newChar = oldChar;
3933 // ignore insets and don't play with deleted text!
3934 if (oldChar != META_INSET && !isDeleted(pos)) {
3936 case text_lowercase:
3937 newChar = lowercase(oldChar);
3939 case text_capitalization:
3941 newChar = uppercase(oldChar);
3945 case text_uppercase:
3946 newChar = uppercase(oldChar);
3951 if (isWordSeparator(pos) || isDeleted(pos)) {
3952 // permit capitalization again
3956 if (oldChar != newChar) {
3957 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3958 if (pos != right - 1)
3960 // step behind the changing area
3964 int erasePos = pos - changes.size();
3965 for (size_t i = 0; i < changes.size(); i++) {
3966 insertChar(pos, changes[i].first,
3969 if (!eraseChar(erasePos, trackChanges)) {
3972 ++right; // expand selection
3980 int Paragraph::find(docstring const & str, bool cs, bool mw,
3981 pos_type start_pos, bool del) const
3983 pos_type pos = start_pos;
3984 int const strsize = str.length();
3986 pos_type const parsize = d->text_.size();
3987 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3988 // Ignore "invisible" letters such as ligature breaks
3989 // and hyphenation chars while searching
3990 while (pos < parsize - 1 && isInset(pos)) {
3991 odocstringstream os;
3992 getInset(pos)->toString(os);
3993 if (!getInset(pos)->isLetter() || !os.str().empty())
3997 if (cs && str[i] != d->text_[pos])
3999 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4001 if (!del && isDeleted(pos))
4008 // if necessary, check whether string matches word
4010 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4013 && !isWordSeparator(pos))
4017 return pos - start_pos;
4021 char_type Paragraph::getChar(pos_type pos) const
4023 return d->text_[pos];
4027 pos_type Paragraph::size() const
4029 return d->text_.size();
4033 bool Paragraph::empty() const
4035 return d->text_.empty();
4039 bool Paragraph::isInset(pos_type pos) const
4041 return d->text_[pos] == META_INSET;
4045 bool Paragraph::isSeparator(pos_type pos) const
4047 //FIXME: Are we sure this can be the only separator?
4048 return d->text_[pos] == ' ';
4052 void Paragraph::deregisterWords()
4054 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4055 Private::LangWordsMap::const_iterator ite = d->words_.end();
4056 for (; itl != ite; ++itl) {
4057 WordList & wl = theWordList(itl->first);
4058 Private::Words::const_iterator it = (itl->second).begin();
4059 Private::Words::const_iterator et = (itl->second).end();
4060 for (; it != et; ++it)
4067 void Paragraph::locateWord(pos_type & from, pos_type & to,
4068 word_location const loc, bool const ignore_deleted) const
4071 case WHOLE_WORD_STRICT:
4072 if (from == 0 || from == size()
4073 || isWordSeparator(from, ignore_deleted)
4074 || isWordSeparator(from - 1, ignore_deleted)) {
4081 // If we are already at the beginning of a word, do nothing
4082 if (!from || isWordSeparator(from - 1, ignore_deleted))
4087 // always move the cursor to the beginning of previous word
4088 while (from && !isWordSeparator(from - 1, ignore_deleted))
4092 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4095 // no need to move the 'from' cursor
4099 while (to < size() && !isWordSeparator(to, ignore_deleted))
4104 void Paragraph::collectWords()
4106 for (pos_type pos = 0; pos < size(); ++pos) {
4107 if (isWordSeparator(pos))
4109 pos_type from = pos;
4110 locateWord(from, pos, WHOLE_WORD);
4111 // Work around MSVC warning: The statement
4112 // if (pos < from + lyxrc.completion_minlength)
4113 // triggers a signed vs. unsigned warning.
4114 // I don't know why this happens, it could be a MSVC bug, or
4115 // related to LLP64 (windows) vs. LP64 (unix) programming
4116 // model, or the C++ standard might be ambigous in the section
4117 // defining the "usual arithmetic conversions". However, using
4118 // a temporary variable is safe and works on all compilers.
4119 pos_type const endpos = from + lyxrc.completion_minlength;
4122 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4123 if (cit == d->fontlist_.end())
4125 Language const * lang = cit->font().language();
4126 docstring const word = asString(from, pos, AS_STR_NONE);
4127 d->words_[lang->lang()].insert(word);
4132 void Paragraph::registerWords()
4134 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4135 Private::LangWordsMap::const_iterator ite = d->words_.end();
4136 for (; itl != ite; ++itl) {
4137 WordList & wl = theWordList(itl->first);
4138 Private::Words::const_iterator it = (itl->second).begin();
4139 Private::Words::const_iterator et = (itl->second).end();
4140 for (; it != et; ++it)
4146 void Paragraph::updateWords()
4154 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4156 SkipPositionsIterator begin = skips.begin();
4157 SkipPositions::iterator end = skips.end();
4158 if (pos > 0 && begin < end) {
4160 if (end->last == pos - 1) {
4165 skips.insert(end, FontSpan(pos, pos));
4169 Language * Paragraph::Private::locateSpellRange(
4170 pos_type & from, pos_type & to,
4171 SkipPositions & skips) const
4173 // skip leading white space
4174 while (from < to && owner_->isWordSeparator(from))
4176 // don't check empty range
4179 // get current language
4180 Language * lang = getSpellLanguage(from);
4181 pos_type last = from;
4182 bool samelang = true;
4183 bool sameinset = true;
4184 while (last < to && samelang && sameinset) {
4185 // hop to end of word
4186 while (last < to && !owner_->isWordSeparator(last)) {
4187 if (owner_->getInset(last)) {
4188 appendSkipPosition(skips, last);
4189 } else if (owner_->isDeleted(last)) {
4190 appendSkipPosition(skips, last);
4194 // hop to next word while checking for insets
4195 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4196 if (Inset const * inset = owner_->getInset(last))
4197 sameinset = inset->isChar() && inset->isLetter();
4198 if (sameinset && owner_->isDeleted(last)) {
4199 appendSkipPosition(skips, last);
4204 if (sameinset && last < to) {
4205 // now check for language change
4206 samelang = lang == getSpellLanguage(last);
4209 // if language change detected backstep is needed
4217 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4220 const_cast<Language *>(owner_->getFontSettings(
4221 inset_owner_->buffer().params(), from).language());
4222 if (lang == inset_owner_->buffer().params().language
4223 && !lyxrc.spellchecker_alt_lang.empty()) {
4225 string const lang_variety =
4226 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4227 lang->setCode(lang_code);
4228 lang->setVariety(lang_variety);
4234 void Paragraph::requestSpellCheck(pos_type pos)
4236 d->requestSpellCheck(pos);
4240 bool Paragraph::needsSpellCheck() const
4242 SpellChecker::ChangeNumber speller_change_number = 0;
4243 if (theSpellChecker())
4244 speller_change_number = theSpellChecker()->changeNumber();
4245 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4246 d->speller_state_.needsCompleteRefresh(speller_change_number);
4248 return d->needsSpellCheck();
4252 bool Paragraph::Private::ignoreWord(docstring const & word) const
4254 // Ignore words with digits
4255 // FIXME: make this customizable
4256 // (note that some checkers ignore words with digits by default)
4257 docstring::const_iterator cit = word.begin();
4258 docstring::const_iterator const end = word.end();
4259 for (; cit != end; ++cit) {
4260 if (isNumber((*cit)))
4267 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4268 WordLangTuple & wl, docstring_list & suggestions,
4269 bool do_suggestion, bool check_learned) const
4271 SpellChecker::Result result = SpellChecker::WORD_OK;
4272 SpellChecker * speller = theSpellChecker();
4276 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4279 locateWord(from, to, WHOLE_WORD, true);
4280 if (from == to || from >= size())
4283 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4284 Language * lang = d->getSpellLanguage(from);
4286 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4289 wl = WordLangTuple(word, lang);
4294 if (needsSpellCheck() || check_learned) {
4296 if (!d->ignoreWord(word)) {
4297 bool const trailing_dot = to < size() && d->text_[to] == '.';
4298 result = speller->check(wl);
4299 if (SpellChecker::misspelled(result) && trailing_dot) {
4300 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4301 result = speller->check(wl);
4302 if (!SpellChecker::misspelled(result)) {
4303 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4305 from << ".." << to << "]");
4307 // spell check with dot appended failed too
4308 // restore original word/lang value
4309 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4310 wl = WordLangTuple(word, lang);
4314 if (!SpellChecker::misspelled(result)) {
4315 // area up to the begin of the next word is not misspelled
4316 while (end < size() && isWordSeparator(end))
4319 d->setMisspelled(from, end, result);
4321 result = d->speller_state_.getState(from);
4325 suggestions.clear();
4327 if (SpellChecker::misspelled(result)) {
4328 LYXERR(Debug::GUI, "misspelled word: \"" <<
4330 from << ".." << to << "]");
4332 speller->suggest(wl, suggestions);
4338 void Paragraph::anonymize()
4340 // This is a very crude anonymization for now
4341 for (char_type & c : d->text_)
4342 if (isLetterChar(c) || isNumber(c))
4347 void Paragraph::Private::markMisspelledWords(
4348 pos_type const & first, pos_type const & last,
4349 SpellChecker::Result result,
4350 docstring const & word,
4351 SkipPositions const & skips)
4353 if (!SpellChecker::misspelled(result)) {
4354 setMisspelled(first, last, SpellChecker::WORD_OK);
4358 SpellChecker * speller = theSpellChecker();
4359 // locate and enumerate the error positions
4360 int nerrors = speller->numMisspelledWords();
4362 SkipPositionsIterator it = skips.begin();
4363 SkipPositionsIterator et = skips.end();
4364 for (int index = 0; index < nerrors; ++index) {
4367 speller->misspelledWord(index, wstart, wlen);
4368 /// should not happen if speller supports range checks
4369 if (!wlen) continue;
4370 docstring const misspelled = word.substr(wstart, wlen);
4371 wstart += first + numskipped;
4372 if (snext < wstart) {
4373 /// mark the range of correct spelling
4374 numskipped += countSkips(it, et, wstart);
4375 setMisspelled(snext,
4376 wstart - 1, SpellChecker::WORD_OK);
4378 snext = wstart + wlen;
4379 numskipped += countSkips(it, et, snext);
4380 /// mark the range of misspelling
4381 setMisspelled(wstart, snext, result);
4382 LYXERR(Debug::GUI, "misspelled word: \"" <<
4383 misspelled << "\" [" <<
4384 wstart << ".." << (snext-1) << "]");
4387 if (snext <= last) {
4388 /// mark the range of correct spelling at end
4389 setMisspelled(snext, last, SpellChecker::WORD_OK);
4394 void Paragraph::spellCheck() const
4396 SpellChecker * speller = theSpellChecker();
4397 if (!speller || empty() ||!needsSpellCheck())
4401 d->rangeOfSpellCheck(start, endpos);
4402 if (speller->canCheckParagraph()) {
4403 // loop until we leave the range
4404 for (pos_type first = start; first < endpos; ) {
4405 pos_type last = endpos;
4406 Private::SkipPositions skips;
4407 Language * lang = d->locateSpellRange(first, last, skips);
4408 if (first >= endpos)
4410 // start the spell checker on the unit of meaning
4411 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4412 WordLangTuple wl = WordLangTuple(word, lang);
4413 SpellChecker::Result result = word.size() ?
4414 speller->check(wl) : SpellChecker::WORD_OK;
4415 d->markMisspelledWords(first, last, result, word, skips);
4419 static docstring_list suggestions;
4420 pos_type to = endpos;
4421 while (start < endpos) {
4423 spellCheck(start, to, wl, suggestions, false);
4427 d->readySpellCheck();
4431 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4433 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4434 if (result || pos <= 0 || pos > size())
4436 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4437 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4442 string Paragraph::magicLabel() const
4445 ss << "magicparlabel-" << id();